Michael J. Spencer | 22120c4 | 2012-04-03 23:09:22 +0000 | [diff] [blame] | 1 | //===- YAMLBench - Benchmark the YAMLParser implementation ----------------===// |
| 2 | // |
| 3 | // The LLVM Compiler Infrastructure |
| 4 | // |
| 5 | // This file is distributed under the University of Illinois Open Source |
| 6 | // License. See LICENSE.TXT for details. |
| 7 | // |
| 8 | //===----------------------------------------------------------------------===// |
| 9 | // |
David Blaikie | 2a52882 | 2014-04-05 20:28:13 +0000 | [diff] [blame] | 10 | // This program executes the YAMLParser on differently sized YAML texts and |
Michael J. Spencer | 22120c4 | 2012-04-03 23:09:22 +0000 | [diff] [blame] | 11 | // outputs the run time. |
| 12 | // |
| 13 | //===----------------------------------------------------------------------===// |
| 14 | |
| 15 | |
| 16 | #include "llvm/ADT/SmallString.h" |
| 17 | #include "llvm/Support/Casting.h" |
| 18 | #include "llvm/Support/CommandLine.h" |
| 19 | #include "llvm/Support/MemoryBuffer.h" |
Michael J. Spencer | 22120c4 | 2012-04-03 23:09:22 +0000 | [diff] [blame] | 20 | #include "llvm/Support/SourceMgr.h" |
Michael J. Spencer | 22120c4 | 2012-04-03 23:09:22 +0000 | [diff] [blame] | 21 | #include "llvm/Support/Timer.h" |
Alex Lorenz | e4bcfbf | 2015-05-07 18:08:46 +0000 | [diff] [blame^] | 22 | #include "llvm/Support/Process.h" |
Michael J. Spencer | 22120c4 | 2012-04-03 23:09:22 +0000 | [diff] [blame] | 23 | #include "llvm/Support/YAMLParser.h" |
Chandler Carruth | 91d19d8 | 2012-12-04 10:37:14 +0000 | [diff] [blame] | 24 | #include "llvm/Support/raw_ostream.h" |
Rafael Espindola | a6e9c3e | 2014-06-12 17:38:55 +0000 | [diff] [blame] | 25 | #include <system_error> |
Michael J. Spencer | 22120c4 | 2012-04-03 23:09:22 +0000 | [diff] [blame] | 26 | |
| 27 | using namespace llvm; |
| 28 | |
| 29 | static cl::opt<bool> |
| 30 | DumpTokens( "tokens" |
| 31 | , cl::desc("Print the tokenization of the file.") |
| 32 | , cl::init(false) |
| 33 | ); |
| 34 | |
| 35 | static cl::opt<bool> |
| 36 | DumpCanonical( "canonical" |
| 37 | , cl::desc("Print the canonical YAML for this file.") |
| 38 | , cl::init(false) |
| 39 | ); |
| 40 | |
| 41 | static cl::opt<std::string> |
| 42 | Input(cl::Positional, cl::desc("<input>")); |
| 43 | |
| 44 | static cl::opt<bool> |
| 45 | Verify( "verify" |
| 46 | , cl::desc( |
| 47 | "Run a quick verification useful for regression testing") |
| 48 | , cl::init(false) |
| 49 | ); |
| 50 | |
| 51 | static cl::opt<unsigned> |
| 52 | MemoryLimitMB("memory-limit", cl::desc( |
| 53 | "Do not use more megabytes of memory"), |
| 54 | cl::init(1000)); |
| 55 | |
Alex Lorenz | e4bcfbf | 2015-05-07 18:08:46 +0000 | [diff] [blame^] | 56 | cl::opt<cl::boolOrDefault> |
| 57 | UseColor("use-color", cl::desc("Emit colored output (default=autodetect)"), |
| 58 | cl::init(cl::BOU_UNSET)); |
| 59 | |
Michael J. Spencer | 22120c4 | 2012-04-03 23:09:22 +0000 | [diff] [blame] | 60 | struct indent { |
| 61 | unsigned distance; |
| 62 | indent(unsigned d) : distance(d) {} |
| 63 | }; |
| 64 | |
| 65 | static raw_ostream &operator <<(raw_ostream &os, const indent &in) { |
| 66 | for (unsigned i = 0; i < in.distance; ++i) |
| 67 | os << " "; |
| 68 | return os; |
| 69 | } |
| 70 | |
Michael J. Spencer | c064a9a | 2013-10-18 22:38:04 +0000 | [diff] [blame] | 71 | /// \brief Pretty print a tag by replacing tag:yaml.org,2002: with !!. |
| 72 | static std::string prettyTag(yaml::Node *N) { |
| 73 | std::string Tag = N->getVerbatimTag(); |
| 74 | if (StringRef(Tag).startswith("tag:yaml.org,2002:")) { |
| 75 | std::string Ret = "!!"; |
| 76 | Ret += StringRef(Tag).substr(18); |
Benjamin Kramer | 548e8a4 | 2015-05-01 15:16:11 +0000 | [diff] [blame] | 77 | return Ret; |
Michael J. Spencer | c064a9a | 2013-10-18 22:38:04 +0000 | [diff] [blame] | 78 | } |
| 79 | std::string Ret = "!<"; |
| 80 | Ret += Tag; |
| 81 | Ret += ">"; |
| 82 | return Ret; |
| 83 | } |
| 84 | |
Michael J. Spencer | 22120c4 | 2012-04-03 23:09:22 +0000 | [diff] [blame] | 85 | static void dumpNode( yaml::Node *n |
| 86 | , unsigned Indent = 0 |
| 87 | , bool SuppressFirstIndent = false) { |
| 88 | if (!n) |
| 89 | return; |
| 90 | if (!SuppressFirstIndent) |
| 91 | outs() << indent(Indent); |
| 92 | StringRef Anchor = n->getAnchor(); |
| 93 | if (!Anchor.empty()) |
| 94 | outs() << "&" << Anchor << " "; |
| 95 | if (yaml::ScalarNode *sn = dyn_cast<yaml::ScalarNode>(n)) { |
| 96 | SmallString<32> Storage; |
| 97 | StringRef Val = sn->getValue(Storage); |
Michael J. Spencer | c064a9a | 2013-10-18 22:38:04 +0000 | [diff] [blame] | 98 | outs() << prettyTag(n) << " \"" << yaml::escape(Val) << "\""; |
Michael J. Spencer | 22120c4 | 2012-04-03 23:09:22 +0000 | [diff] [blame] | 99 | } else if (yaml::SequenceNode *sn = dyn_cast<yaml::SequenceNode>(n)) { |
Michael J. Spencer | c064a9a | 2013-10-18 22:38:04 +0000 | [diff] [blame] | 100 | outs() << prettyTag(n) << " [\n"; |
Michael J. Spencer | 22120c4 | 2012-04-03 23:09:22 +0000 | [diff] [blame] | 101 | ++Indent; |
| 102 | for (yaml::SequenceNode::iterator i = sn->begin(), e = sn->end(); |
| 103 | i != e; ++i) { |
| 104 | dumpNode(i, Indent); |
| 105 | outs() << ",\n"; |
| 106 | } |
| 107 | --Indent; |
| 108 | outs() << indent(Indent) << "]"; |
| 109 | } else if (yaml::MappingNode *mn = dyn_cast<yaml::MappingNode>(n)) { |
Michael J. Spencer | c064a9a | 2013-10-18 22:38:04 +0000 | [diff] [blame] | 110 | outs() << prettyTag(n) << " {\n"; |
Michael J. Spencer | 22120c4 | 2012-04-03 23:09:22 +0000 | [diff] [blame] | 111 | ++Indent; |
| 112 | for (yaml::MappingNode::iterator i = mn->begin(), e = mn->end(); |
| 113 | i != e; ++i) { |
| 114 | outs() << indent(Indent) << "? "; |
| 115 | dumpNode(i->getKey(), Indent, true); |
| 116 | outs() << "\n"; |
| 117 | outs() << indent(Indent) << ": "; |
| 118 | dumpNode(i->getValue(), Indent, true); |
| 119 | outs() << ",\n"; |
| 120 | } |
| 121 | --Indent; |
| 122 | outs() << indent(Indent) << "}"; |
| 123 | } else if (yaml::AliasNode *an = dyn_cast<yaml::AliasNode>(n)){ |
| 124 | outs() << "*" << an->getName(); |
Benjamin Kramer | 619c4e5 | 2015-04-10 11:24:51 +0000 | [diff] [blame] | 125 | } else if (isa<yaml::NullNode>(n)) { |
Michael J. Spencer | c064a9a | 2013-10-18 22:38:04 +0000 | [diff] [blame] | 126 | outs() << prettyTag(n) << " null"; |
Michael J. Spencer | 22120c4 | 2012-04-03 23:09:22 +0000 | [diff] [blame] | 127 | } |
| 128 | } |
| 129 | |
| 130 | static void dumpStream(yaml::Stream &stream) { |
| 131 | for (yaml::document_iterator di = stream.begin(), de = stream.end(); di != de; |
| 132 | ++di) { |
| 133 | outs() << "%YAML 1.2\n" |
| 134 | << "---\n"; |
| 135 | yaml::Node *n = di->getRoot(); |
| 136 | if (n) |
| 137 | dumpNode(n); |
| 138 | else |
| 139 | break; |
| 140 | outs() << "\n...\n"; |
| 141 | } |
| 142 | } |
| 143 | |
| 144 | static void benchmark( llvm::TimerGroup &Group |
| 145 | , llvm::StringRef Name |
| 146 | , llvm::StringRef JSONText) { |
| 147 | llvm::Timer BaseLine((Name + ": Loop").str(), Group); |
| 148 | BaseLine.startTimer(); |
| 149 | char C = 0; |
| 150 | for (llvm::StringRef::iterator I = JSONText.begin(), |
| 151 | E = JSONText.end(); |
| 152 | I != E; ++I) { C += *I; } |
| 153 | BaseLine.stopTimer(); |
| 154 | volatile char DontOptimizeOut = C; (void)DontOptimizeOut; |
| 155 | |
| 156 | llvm::Timer Tokenizing((Name + ": Tokenizing").str(), Group); |
| 157 | Tokenizing.startTimer(); |
| 158 | { |
| 159 | yaml::scanTokens(JSONText); |
| 160 | } |
| 161 | Tokenizing.stopTimer(); |
| 162 | |
| 163 | llvm::Timer Parsing((Name + ": Parsing").str(), Group); |
| 164 | Parsing.startTimer(); |
| 165 | { |
| 166 | llvm::SourceMgr SM; |
| 167 | llvm::yaml::Stream stream(JSONText, SM); |
| 168 | stream.skip(); |
| 169 | } |
| 170 | Parsing.stopTimer(); |
| 171 | } |
| 172 | |
| 173 | static std::string createJSONText(size_t MemoryMB, unsigned ValueSize) { |
Alp Toker | e69170a | 2014-06-26 22:52:05 +0000 | [diff] [blame] | 174 | std::string JSONText; |
| 175 | llvm::raw_string_ostream Stream(JSONText); |
| 176 | Stream << "[\n"; |
Michael J. Spencer | 22120c4 | 2012-04-03 23:09:22 +0000 | [diff] [blame] | 177 | size_t MemoryBytes = MemoryMB * 1024 * 1024; |
Alp Toker | e69170a | 2014-06-26 22:52:05 +0000 | [diff] [blame] | 178 | while (JSONText.size() < MemoryBytes) { |
| 179 | Stream << " {\n" |
| 180 | << " \"key1\": \"" << std::string(ValueSize, '*') << "\",\n" |
| 181 | << " \"key2\": \"" << std::string(ValueSize, '*') << "\",\n" |
| 182 | << " \"key3\": \"" << std::string(ValueSize, '*') << "\"\n" |
| 183 | << " }"; |
| 184 | Stream.flush(); |
| 185 | if (JSONText.size() < MemoryBytes) Stream << ","; |
| 186 | Stream << "\n"; |
Michael J. Spencer | 22120c4 | 2012-04-03 23:09:22 +0000 | [diff] [blame] | 187 | } |
Alp Toker | e69170a | 2014-06-26 22:52:05 +0000 | [diff] [blame] | 188 | Stream << "]\n"; |
| 189 | Stream.flush(); |
| 190 | return JSONText; |
Michael J. Spencer | 22120c4 | 2012-04-03 23:09:22 +0000 | [diff] [blame] | 191 | } |
| 192 | |
| 193 | int main(int argc, char **argv) { |
| 194 | llvm::cl::ParseCommandLineOptions(argc, argv); |
Alex Lorenz | e4bcfbf | 2015-05-07 18:08:46 +0000 | [diff] [blame^] | 195 | bool ShowColors = UseColor == cl::BOU_UNSET |
| 196 | ? sys::Process::StandardOutHasColors() |
| 197 | : UseColor == cl::BOU_TRUE; |
Michael J. Spencer | 22120c4 | 2012-04-03 23:09:22 +0000 | [diff] [blame] | 198 | if (Input.getNumOccurrences()) { |
Rafael Espindola | adf21f2 | 2014-07-06 17:43:13 +0000 | [diff] [blame] | 199 | ErrorOr<std::unique_ptr<MemoryBuffer>> BufOrErr = |
| 200 | MemoryBuffer::getFileOrSTDIN(Input); |
| 201 | if (!BufOrErr) |
Michael J. Spencer | 22120c4 | 2012-04-03 23:09:22 +0000 | [diff] [blame] | 202 | return 1; |
Rafael Espindola | 3f6481d | 2014-08-01 14:31:55 +0000 | [diff] [blame] | 203 | MemoryBuffer &Buf = *BufOrErr.get(); |
Michael J. Spencer | 22120c4 | 2012-04-03 23:09:22 +0000 | [diff] [blame] | 204 | |
| 205 | llvm::SourceMgr sm; |
| 206 | if (DumpTokens) { |
Rafael Espindola | 3f6481d | 2014-08-01 14:31:55 +0000 | [diff] [blame] | 207 | yaml::dumpTokens(Buf.getBuffer(), outs()); |
Michael J. Spencer | 22120c4 | 2012-04-03 23:09:22 +0000 | [diff] [blame] | 208 | } |
| 209 | |
| 210 | if (DumpCanonical) { |
Alex Lorenz | e4bcfbf | 2015-05-07 18:08:46 +0000 | [diff] [blame^] | 211 | yaml::Stream stream(Buf.getBuffer(), sm, ShowColors); |
Michael J. Spencer | 22120c4 | 2012-04-03 23:09:22 +0000 | [diff] [blame] | 212 | dumpStream(stream); |
Alex Lorenz | e4bcfbf | 2015-05-07 18:08:46 +0000 | [diff] [blame^] | 213 | if (stream.failed()) |
| 214 | return 1; |
Michael J. Spencer | 22120c4 | 2012-04-03 23:09:22 +0000 | [diff] [blame] | 215 | } |
| 216 | } |
| 217 | |
| 218 | if (Verify) { |
| 219 | llvm::TimerGroup Group("YAML parser benchmark"); |
| 220 | benchmark(Group, "Fast", createJSONText(10, 500)); |
| 221 | } else if (!DumpCanonical && !DumpTokens) { |
| 222 | llvm::TimerGroup Group("YAML parser benchmark"); |
| 223 | benchmark(Group, "Small Values", createJSONText(MemoryLimitMB, 5)); |
| 224 | benchmark(Group, "Medium Values", createJSONText(MemoryLimitMB, 500)); |
| 225 | benchmark(Group, "Large Values", createJSONText(MemoryLimitMB, 50000)); |
| 226 | } |
| 227 | |
| 228 | return 0; |
| 229 | } |